Sora, OpenAI’s new generative video app, has been available for barely a week and already sparked a major controversy for the company. As users steer prompts toward recognizable pop-culture aesthetics and famous characters, Sora has produced an outpouring of clips tied to well-known anime and video-game franchises — generating intense debate about the technology’s reach and limits. PC Gamer has chronicled several of the most notable examples.
It appears OpenAI did not deploy strong safeguards to prevent copyrighted works from being used as training or inspiration fodder for Sora. That oversight has produced a backlash significant enough that CEO Sam Altman has publicly signaled forthcoming guardrails and controls for Sora users.
Like ChatGPT, Sora transforms user text prompts into generated output — here, short-form video. Its capabilities range from crafting realistic deepfakes of real people to imitating distinct visual styles. One viral clip shows a hyper-realistic Pikachu being sliced on a grill with a figure resembling OpenAI’s CEO; another recreates a Cyberpunk 2077 mission with details (map layout, vehicle designs, voice qualities) that closely mirror the original game. The resemblance in some of these outputs has alarmed players and creators alike. A widely shared short showing the Pikachu scene has circulated across platforms.
Observers warn that Sora’s realism could fuel misinformation and pose security risks when footage convincingly depicts public figures. Critics are also asking how Sora achieves such fidelity and what datasets underpin its outputs. According to reporting in the Wall Street Journal, OpenAI has begun asking copyright holders to opt out of Sora’s models. Some companies have taken advantage of that option, and OpenAI appears to have instituted stricter blocks around certain properties — notably many Marvel and Disney assets. The Wall Street Journal provides more detail.
By contrast, a number of Japanese entertainment properties seem less consistently restricted by Sora’s current safeguards. The app can produce visuals that very closely mimic anime such as Attack on Titan, Dragon Ball Z, and One Piece, a reality that has drawn public criticism. Social posts criticizing the trend have amplified the debate, and Japanese politician Akihisa Shiozaki reportedly convened an urgent meeting with government officials to discuss intellectual-property responses, with attendees urging swift action to protect cultural assets.
In a blog update, Altman acknowledged that Sora’s early release has produced clearer, immediate feedback on how people actually use the product. He said OpenAI is working to give rights holders more precise controls — beyond a simple opt-in or opt-out — so they can define when and how their characters and properties are used. The company expects to iterate on these systems as edge cases and gaps are discovered. Altman’s full post has additional context.
A recurring flashpoint is Nintendo, whose characters frequently inspire Sora prompts. AI-generated clips of a person resembling Altman saying “I hope Nintendo doesn’t sue us” have circulated widely, fueling discussion about how far generative tools can push copyrighted material. While Nintendo denied reports that it was coordinating with the Japanese government to block AI usage of its IP, the company reiterated that it will continue to take necessary action against any infringement of its intellectual-property rights. Video Games Chronicle summarizes Nintendo’s response, and Nintendo’s own social post restated its stance. Nintendo’s statement on X.
Source: Polygon
